Method Pack . M9 . Interactive

Building an MEL System from Scratch

A 3-hour interactive Practice Pack. You have joined a programme that has been running for two years with no formal monitoring. Walk through six months of setup and walk out with a 90-day MEL plan you can actually execute.

4 modules ~3 hours Interactive India-context
Your progress
0% complete
Your Capstone

90-Day MEL Setup Plan

Walk in with an organisation that has no formal MEL. Walk out with a phased 90-day plan covering audit findings, theory of change, indicator framework, data collection tools, and learning-loop design. Built automatically from your module answers.

Module 1 . ~25 min

Month 1: Audit what exists

Every organisation already monitors something, even if nobody calls it MEL. The first month is about discovering what already exists, what is being collected informally, and where the gaps actually are. Resist the temptation to design a new system before understanding the old one.

The five places data already lives

The audit conversation

Sit with three people individually: (a) the programme director, (b) a senior field coordinator, and (c) a data-entry person if one exists. Ask each of them:

  1. What information do you collect regularly? (probe: registers, forms, apps)
  2. What do you do with it after collecting? (probe: who reads it, what decisions does it feed?)
  3. What do you wish you knew but currently do not?
  4. What reporting burden do you find pointless?
Worked example

Gram Vikas, an NRLM-linked livelihoods programme in Odisha, had been running for three years with no MEL officer. The audit found: (1) SHG meeting registers with attendance and savings data in paper form across 120 villages. (2) Monthly field reports submitted by area coordinators as Word documents with inconsistent formats. (3) NRLM MIS data entry happening at the block level with a two-month lag. (4) Zero outcome-level data.

Key insight: The SHG registers were the richest dataset -- just never aggregated. The new MEL system started by digitising those, not by creating a new survey instrument.

Your MEL Audit Sheet

Fill these for your organisation. Your answers save automatically and flow into the final capstone.

e.g., "Gram Vikas -- NRLM-linked SHG livelihoods programme, 120 villages, Odisha, 3 years running"
Registers, MIS, reports, WhatsApp groups, Excel sheets, government reporting
e.g., NRLM MIS, UDISE+, HMIS, PFMS, Awaas+, PM-KISAN
Saved
Self-check
You join a nutrition programme in Jharkhand that has been running for two years. The team says "we have no data." What is your most productive first step?
Design a new baseline survey
Hire a data entry operator
Ask to see funder reports, field registers, and the ICDS/Poshan Tracker data they already submit
Set up a KoboToolbox account
Correct. "We have no data" almost never means zero data exists. It means nobody has aggregated or used it. The audit conversation reveals what already exists before you add new collection burden.
Module 2 . ~30 min

Month 2: Theory of Change + Indicator Framework

The theory of change is not a wall poster. It is the backbone of your indicator framework. Every indicator you select must connect to a specific causal link in the ToC. If an indicator does not test a causal assumption, it is a vanity metric.

Building the ToC in practice

Run a half-day workshop with the programme team. Use this sequence:

  1. Start from the right -- what is the long-term change this programme seeks? (e.g., "Women in 120 villages have sustained increase in household income and economic agency.")
  2. Work backwards -- what intermediate outcomes must be true for that to happen? (e.g., "SHG members access formal credit," "Micro-enterprises survive beyond 12 months.")
  3. Identify outputs -- what does the programme directly deliver? (e.g., "SHG formation and training," "Enterprise development support.")
  4. Map assumptions -- what must be true between each link? (e.g., "Banks will lend to SHGs," "Market demand exists for the products.")

From ToC to indicators: the three-tier framework

TierWhat it measuresFrequencySource
Tier 1: Activity trackingInputs and outputs (training conducted, SHGs formed, loans disbursed)MonthlyMIS, field reports
Tier 2: Outcome monitoringIntermediate changes (repayment rates, enterprise survival, income change)QuarterlyStructured follow-up, government data
Tier 3: Impact signalsLong-term change (economic agency, food security, asset ownership)Annual or evaluation-onlySurveys, qualitative deep-dives

The "20 indicator" ceiling

Indian development programmes routinely operate with 60-80 indicators because every funder adds their own. Effective MEL systems cap at 15-20 actively monitored indicators. The rest stay in the logframe but are not part of the routine data collection. If your field team cannot explain what each indicator means and why it matters, cut it.

Government scheme alignment

If your programme operates under NRLM, PMAY-G, Samagra Shiksha, or NHM, the government scheme already has a results framework. Do not duplicate it. Align your Tier 1 indicators to the scheme's reporting requirements and add only the Tier 2-3 indicators that the scheme framework misses.

Your ToC and Indicator Framework

Build the skeleton. These flow into your capstone.

What must be true for your ToC to hold? List 2-3 testable assumptions.
Saved
Self-check
Your indicator framework has 52 indicators. The field team can reliably collect 18 per month. What is the best response?
Hire more data collectors
Prioritise 15-18 for routine monitoring, move the rest to periodic or evaluation-only collection
Automate all 52 using a mobile app
Report all 52 but only verify a sample
Correct. Data quality collapses when collection burden exceeds capacity. The solution is tiering, not adding staff or technology. Technology helps with the 18 you keep, not with justifying 52 nobody reads.
Module 3 . ~30 min

Month 3: Data Collection Systems + Tools

This is where most MEL setup projects stall. The team buys a tool (KoboToolbox, CommCare, ODK, SurveyCTO) before designing the data flow. The tool is the last decision, not the first.

Design the data flow first

  1. Who collects? Field coordinators? Community volunteers? Beneficiaries themselves? Each has different capacity, literacy, and motivation levels.
  2. When? Real-time (during activity), weekly aggregate, or monthly reporting? Over-frequent collection kills compliance. Under-frequent collection loses the signal.
  3. How does it move? Paper to block office to Excel? Mobile app to cloud? WhatsApp photo to someone who enters it? Map the actual flow, not the aspirational one.
  4. Who cleans and verifies? This is the step that is always under-resourced. Budget 2-3 days per month for a data person to run validation checks.
  5. Who reads the dashboard? If nobody reads the output, the system will die within six months regardless of the technology.

Tool selection matrix

ToolBest forLimitationsCost (2026)
KoboToolboxSurvey-style data, offline collection, moderate complexityWeak on longitudinal tracking, no built-in case managementFree (OCHA-hosted)
CommCareCase management, health/nutrition, repeat visitsSteeper learning curve, costs at scaleFree up to 50 users, then $150+/mo
Google Sheets + AppSheetTiny teams, rapid prototyping, zero budgetBreaks above 500 rows, no offline, no validationFree
DHIS2Health system data, government alignmentHeavy setup, needs technical supportFree (open source)
Custom Excel + paperWhen digital literacy is very lowError-prone, no real-time visibilityStaff time only
Worked example

Gram Vikas (continued): After the audit, the MEL officer chose KoboToolbox for monthly SHG data (savings, attendance, loan repayment) collected by area coordinators on Android phones. Paper SHG registers remained as the primary record; KoboToolbox digitised a monthly summary. A Google Sheet dashboard was shared with the programme director every Monday. Total setup cost: Rs 0 (KoboToolbox free tier) + Rs 45,000 for coordinator training (2 days, 15 coordinators).

Your Data Collection Design

Design the data flow for your programme. These flow into your capstone.

Tier 1 (activities): ___. Tier 2 (outcomes): ___. Tier 3 (impact): ___.
Saved
Self-check
A programme director says: "We need CommCare because it is what NITI Aayog recommends." Your programme is a small education NGO with 8 field staff and no IT person. What is the right response?
Set up CommCare immediately since it has government endorsement
Build a custom app instead
Start with KoboToolbox or Google Sheets, which match your team's capacity, and migrate later if needed
Hire an IT consultant to manage CommCare
Correct. Tool choice must match team capacity, not prestige. CommCare is excellent for case management at scale, but an 8-person education programme without IT support will abandon it within three months. Start simple, prove the data flow works, then upgrade.
Module 4 . ~25 min

Month 6: Learning Loops + Adaptive Management

Data collection without learning loops is bureaucracy. The entire point of an MEL system is to feed decisions. If the data is collected, cleaned, and filed without anyone acting on it, the system is already dead.

The four learning loops

The decision log

The simplest, most powerful tool in adaptive management. A shared Google Doc or register with four columns:

DateWhat the data showedDecision takenFollow-up by when
15 Mar 2026SHG repayment rate dropped from 92% to 78% in Koraput blockField visit to 5 lowest-performing SHGs this week; check if linked to crop failure22 Mar 2026
15 Mar 2026Training attendance above 90% in all blocks except Rayagada (62%)Reschedule Rayagada trainings to post-harvest; pilot evening sessions1 Apr 2026
The funder conversation

Most Indian funders say they want adaptive management but reward rigid compliance. The way to build funder trust in learning loops is to share the decision log proactively. When you show a funder "here is what the data told us, here is what we changed, here is what happened," you build more credibility than a clean logframe ever will.

Your Learning Loop Design

Design the feedback mechanisms. These flow into your capstone.

Proactive decision-log sharing? Quarterly narrative? Annual learning report?
Saved
Self-check
Your MEL system has been running for four months. The dashboard is updated monthly. But you notice the programme team never changes anything based on the data. What is the most likely root cause?
The indicators are wrong
The data quality is poor
There is no structured learning loop -- the data is presented but never discussed in a decision-making forum
The dashboard design is not user-friendly
Correct. The most common failure mode of MEL systems is not bad data or wrong indicators -- it is the absence of a structured forum where data meets decisions. A dashboard nobody discusses is decoration, not management.
Capstone

Your 90-Day MEL Setup Plan

You have completed the four modules. Click Build my brief to compile everything into a single 90-day MEL setup plan. Copy as markdown, print as PDF, or share with your team.

90-Day MEL Setup Plan

Click "Build my brief" -- your module answers will be pulled into the artefact. Edit or refine afterwards if needed.

Your brief will appear here when you click "Build my brief". It will draw from your answers in Modules 1-4 (which are saved in your browser). Empty fields show as placeholders -- you can either go back and fill them, or edit them here directly after building.

Where to go next on ImpactMojo

Done?

Share this 90-day plan with your programme director before the next team meeting. The plan is only as good as the buy-in you build around it.

Help us improve this pack: feedback form.

All Practice Packs →